According to the characteristic of real-time information interaction under auto-driving mode, comprehensively considering the uncertainty of network traffic demand and the balance of traffic flow at the bottleneck of intersections, a bi-level optimization model for dynamic traffic assignment and signal control under auto-driving mode is proposed. In the lower layer, each vehicle travels along the shortest path determined by the OD data obtained from the autopilot system, which reflects the user's optimization, then the traffic demand of each direction at intersections in a certain interval can be predicted. On the premise of meeting the traffic demand, the control mode with minimum delay is selected. The objective of the upper layer is to minimize the average delay of vehicles on a road. The traffic demand of the upper layer is determined by the lower layer, and the dynamic traffic distribution of vehicles in the lower layer is affected by the intersection control mode determined by the upper layer .Combining frank-wolf algorithm and enumeration method, an iterative algorithm is proposed. An adaptive control scheme of free release for autonomous vehicles at signal-free intersection is presented. As a value of control mode in numerical simulation experiment, the capacity of free release is obviously improved compared with other free release schemes. The numerical simulation experiments verify the correctness and feasibility of the model and the algorithm. The results show that the model achieves the flexible choice of cross signal control mode and the optimal allocation of traffic flow, which reduces the delay of vehicles significantly, and it improves the efficiency of road and controls the occurrence of intersection supersaturation effectively.

CICTP